alansaber

8 ms·

271.
by alansaber 3mo ago
Sounds logical to me. It's already remarkable that we haven't done this with fission weapons. It's not unreasonable to assume that within a century we'll have antimatter weapons. Destructive potential only goes up.
272.
by alansaber 3mo ago
SLM language filters might actually be a good application of AI? As well as very sparing regex for bad words.
273.
Bannerlord Coop(steamcommunity.com)
2 points by alansaber 3mo ago|0 comments
274.
Towards Domain-Specific Compaction(lexifina.com)
1 points by alansaber 3mo ago|0 comments
275.
by alansaber 3mo ago
TBF gaming benchmarks is not something new to AI
276.
by alansaber 3mo ago
it is definitely more verbose.
277.
by alansaber 3mo ago
So AI is mostly a thin glue between deterministic processes. Doesn't change the fact that that is enough to achieve an extremely large amount of tasks.
278.
by alansaber 3mo ago
What each of these is doing, fundamentally, is solving context management in an opinionated way (that and guardrails).
279.
by alansaber 3mo ago
Sounds like you want something similar to /goal mode in Codex.
280.
by alansaber 3mo ago
I don't find the argument that advanced AI will leave 20% of society permanently unemployed particularly compelling
281.
by alansaber 3mo ago
Yep, a pre-wrapped Codex for enterprise. Seems like the logical evolution from setting up an MCP.
282.
by alansaber 3mo ago
So, a security-oriented cloud agent framework? Better call it an OS.
283.
by alansaber 3mo ago
Awesome.
284.
by alansaber 3mo ago
We probably want to discourage frontier labs from security testing in prod
285.
by alansaber 3mo ago
The P in MCP stands for REST
286.
by alansaber 3mo ago
Skills definitely can pollute the context window. They can be verbose. You can have a very long list of advertised skills. You can try and hack around with skill execution inside of subagents etc but there is always a trade-off.
287.
by alansaber 3mo ago
This is the new enterprise trend (RAG? no, we need our own pi plugin/adaptation).
288.
by alansaber 3mo ago
That's hilarious
289.
by alansaber 3mo ago
Feels like a problem where you have to quantify/test for your own use-case. Otherwise just go by the generic SWE-bench scores etc.
290.
by alansaber 3mo ago
It doesn't do anything unique WRT context management. It's just much less opinionated.
291.
Our top agent latency optimisations(lexifina.com)
2 points by alansaber 3mo ago|0 comments
292.
Cloud Agents Should Use (Some) Code Execution(lexifina.com)
2 points by alansaber 3mo ago|0 comments
293.
by alansaber 3mo ago
Obsessing over the medium of work rather than the work itself has been a subculture since forever
294.
by alansaber 3mo ago
Recent models have had a worrying lean towards longer output. If you're going to serve up some AI slop, at least distil it to the smallest possible footprint possible.
295.
by alansaber 3mo ago
Yep. Though hard to write a compelling blog by being to-the-point without data
296.
by alansaber 3mo ago
Some interesting concepts, but reads like satire.
297.
by alansaber 3mo ago
Sounds just like using the gym for therapy
298.
by alansaber 3mo ago
"Sometimes the asset itself should become the center of the workspace, with the delegated work happening around it." Arguably always, for anything details-oriented. You can get away with coding from TUI, but video editing etc? Hel
299.
by alansaber 3mo ago
I think there is so much abstraction/simplification you can make when fundamentally you need a strong VCS underwriting everything
300.
by alansaber 3mo ago
Are you using codex app? In CLI you can fork/rename chats. Albeit you cant fork rename in one action which is annoying. I find naming chats, putting chats into folders, giving chat tabs unique colors according to their purpose helps me

More ›